Analysis: A worker with a workplace issue can use AI to receive an assessment of their rights, a grievance letter and a draft complaint form within minutes
Most people who believe they have been treated wrongly at work do nothing about it. In the largest survey of individual workplace conflict ever conducted in Britain, fewer than one in 10 workers who experienced conflict took formal action of any kind, and fewer than one in 200 brought a tribunal claim. Researchers who study employment disputes describe this as a pyramid, with everyone holding a legitimate grievance at the base and the tiny fraction who ever follow one up at the apex.
People give up for reasons that have little to do with the perceived strength of their case. Fear of retaliation is part of it, so too is the cost of advice, or not knowing what the law actually entitles you to, and many people simply decide that grievance procedures are for other kinds of workers.
Whatever the reason, HR departments, tribunals and commissions have long assumed that most workplace problems never reach them. That assumption is now being tested. A worker who suspects they have been treated unlawfully can describe the situation to an AI model and receive an assessment of their rights, a structured grievance letter and a draft complaint form within minutes and at no cost.
We need your consent to load this rte-player contentWe use rte-player to manage extra content that can set cookies on your device and collect data about your activity. Please review their details and accept them to load the content.Manage Preferences
From RTÉ Brainstorm, why workers don’t like their company’s HR departments
Employment lawyers in the UK estimate that AI-drafted content now features in virtually every tribunal case they see, while firms acting for employers describe grievances quoting legal principles at length. The situation in Ireland is similar. The Workplace Relations Commission (WRC) received just over 10,500 complaint applications in 2025, a rise of 44% on the previous year, while the number of decisions it issued fell.
The Law Society believe AI is one contributor to that pressure, leading to the WRC publishing guidance for litigants, prompted partly by submissions citing case law that turned out to be inaccurate or invented. Whatever is submitted to the WRC is treated as the complainant’s own work, whether AI helped to produce it or not.
Making a complaint is a right
Where the use of AI in workplace disputes has been discussed, it has primarily been looked at from the employer’s perspective, and mostly as a problem to be managed. That framing is worth examining. The right to raise a grievance with your employer, and to bring a complaint to the WRC or Labour Court, is itself an employment right, as much a part of the relationship as the right to be paid for your work. Sociologists of work have long argued that the employment relationship is asymmetric by design. Employers write the contracts, set the grievance and disciplinary procedures and hold the records that a dispute will turn on, and they can pay for professional representation when one arises.
We need your consent to load this rte-player contentWe use rte-player to manage extra content that can set cookies on your device and collect data about your activity. Please review their details and accept them to load the content.Manage Preferences
From RTÉ Brainstorm, what goes on at the Workplace Relations Commission?
Employees are left relying on whatever advice they can find or afford at the point where they may be least able to afford it. Research following workers through their disputes found that too few claims, if anything, go forward rather than too many. This does not make employer concerns imaginary, as fabricated citations and inflated expectations generate costs for both parties and serve nobody, least of all the people who bring them. It does, however, mean that a rise in complaints should not be read, on its own, as an abuse of the system.
Some of what is now reaching HR departments and the WRC is the exercise of rights that always existed, by people who previously may have lacked the means to exercise them. It is also worth highlighting that when organisations use AI to screen, recruit and manage staff, it is generally framed using language describing productivity and efficiency gains, but the description shifts to one of risk and crisis when workers use the same technologies to assert their rights.
The machine that never says no
Workers arguably have more to lose than employers from using this technology badly, and the biggest risk is sycophancy. AI models are trained partly on human feedback and, because people reward agreement, the systems learn to agree. A recent study found that many of the most prominent models affirmed users’ viewpoints 49% more often than humans did, and that people advised by these agreeable systems became more convinced they were right. If, for example, you describe a conflict with your manager to an AI model, it will generally accept your account, attach legal labels to the behaviour and estimate what a claim might be worth.
We need your consent to load this rte-player contentWe use rte-player to manage extra content that can set cookies on your device and collect data about your activity. Please review their details and accept them to load the content.Manage Preferences
From RTÉ Radio 1’s Saturday with Cormac Ó Headhra, how worried should we be about AI replacing jobs?
Part of what a paid (human) adviser traditionally provides is a dispassionate assessment of whether a case is worth pursuing. A system tuned toward agreement provides a polished draft of a grievance without that judgement, and practitioners describe complainants pursuing draining proceedings on the strength of settlement expectations generated by an AI model that never examined the evidence.
The consequences are evident in a case reported earlier this year, where a former car wash employee lost his WRC complaint after accepting that the form had been prepared on his behalf and its narrative generated using AI, leaving him unable to clearly explain aspects of a complaint put forward in his own name. The adjudicator found that, while AI assistance is not improper, complainants remain responsible for the accuracy of what is submitted, and his reliance on unverified material materially reduced the weight given to his evidence.
How to use AI without weakening your case
For anyone considering raising a workplace-related grievance, find out first of all how your workplace handles disputes rather than consulting ChatGPT. Most employers have a grievance procedure, and adjudicators generally expect it to have been used before a complaint reaches the WRC.
We need your consent to load this rte-player contentWe use rte-player to manage extra content that can set cookies on your device and collect data about your activity. Please review their details and accept them to load the content.Manage Preferences
From RTÉ Radio 1’s Brendan O’Connor Show, how to ask for a pay rise
Read it and note what it asks you to do first, because it will often ask you to raise the matter informally with a manager before anything formal begins. That step can resolve many issues, and a record of having tried is important if matters go further. If you work in a unionised organisation, talk to your shop steward or union official early, as they deal with these procedures regularly and can represent you through the process. None of this requires AI.
Where AI may help is in preparation, and there it works best when used deliberately rather than passively, starting with facts rather than law. AI sysems are adept at organising material, so use it to build a chronology and to assemble your emails, rosters and payslips into a coherent account, while treating its statements of Irish employment law with far more caution, as they regularly misinterpret its nuances.
Instruct it to challenge you. Ask what the employer’s strongest response would be, where your case is weakest, and tell it directly not to reassure you. Check every citation yourself, as legislation is freely available on the Irish Statute Book and WRC decisions are published on the Commission’s website. If you cannot find a case cited by AI, assume it does not exist.
We need your consent to load this rte-player contentWe use rte-player to manage extra content that can set cookies on your device and collect data about your activity. Please review their details and accept them to load the content.Manage Preferences
From RTÉ Radio 1’s Oliver Callan Show, should you use AI to write a CV?
Watch the timeframe too, as most WRC complaints must be lodged within six months of the matter complained of, extendable to twelve only where reasonable cause is shown. Avoid pasting colleagues’ personal information into consumer AI tools, and remember that free advice with a human on the other end is available through the WRC’s information service, Citizens Information and Free Legal Advice Centres.
Above all, keep your own voice in the final document, because you will have to stand over every sentence of it in a hearing room. None of this is a reason to hold back from a complaint that deserves pursuing. Making one is your right. What is at stake in using AI carelessly is not the technology’s credibility but yours, in a situation where your account of events is the evidence and your ability to explain it is what matters.
Follow RTÉ Brainstorm on WhatsApp and Instagram for more stories and updates
The views expressed here are those of the author and do not represent or reflect the views of RTÉ
